Comparative Analysis between Iterative Threshold Technique and Otsu’s Method

Abstract: This paper compares and examines the two image segmentation techniques, traditional Otsu method and the iterative Otsu method which includes the new technique called the iterative tri-class threshold technique. The iterative method selects the first threshold value using the Otsu�s threshold and then computes the mean values of the two classes as separated by this threshold. Using this threshold value and the two mean values as calculated, this method separates the image into three classes instead of dividing into two classes as is done in the standard Otsu�s method. The first class is the background and the second class is the foreground. The third class is the region that needs to be processed in the next iteration. This paper focuses on two parts, at first on stopping criteria and number of iteration in iterative method. Second, comparison between these two methods based on results.

Keywords: Otsu�s method [6], segmentation, threshold, tri-class threshold technique [1].